New 2GHz touchscreen scope from Tek, June 6th

[rf-design]

I take a closer look at the chip plot. It show some layers below the top-metal which contain the pads for the high pincount ball array.

The board photo indicate one chip for each channel. But from the plot 8 differential analog inputs could be identified and 4 ADCs groups each with a possible interleave factor of 16. In addition each chip contain 18 complementary digital inputs.

My guess is that this Tek049 chip is designed to be applied also for low-cost instruments where each Tek049 could serve minimum 4 analog channels and 16+2 digital channels. So the Tek049 seems to be a platform chip for the 5-series and everything below.

What is your thought?

[EEVblog]

My guess is that this Tek049 chip is designed to be applied also for low-cost instruments where each Tek049 could serve minimum 4 analog channels and 16+2 digital channels. So the Tek049 seems to be a platform chip for the 5-series and everything below.
What is your thought?

Yep, I noticed that too.
A 12bit 4 channel muxed 500MHz MSO based on this would be nice.
At 6.25GS/s / 4 one chip is still enough sample rate for a 4CH 500MHz front end.

[mikeselectricstuff]

I had a good play with this scope at the EDS show today, and a long chat with couple of UK Tek guys on the stand. In no particular order....

This thing is BIG, really big. Too big for a lot of benches.
The pins on the digital probe are numerous and tiny - I have doubts as to how robust they will be.

There are some aspects of the touch UI that are a bit non-intuitive, in particular double-tap on the icons at the button to bring up a menu instead of touch and hold. I couldn't really see why it shouldn't just bring up the menu on the first touch,  and if there is a reason not to,  touch and hold would be more sensible (cf. Android).
Having to drag the icons to the trash to turn off is horrible, mostly because of the distance you have to drag (Tek guy agreed)- the menu could just have a "go away" icon.

Although the UI looks pretty fast and responsive at first look, once you start turning other channels on, it slows down. With a lot of channels, it slows A LOT. Unless there was some weird setup causing this ( no, decodes weren't on) it was looking pretty terrible with all channels on. It may be that reducing memory would help, though pressing "fast acq" reduced the memory so much that the waveform looked terrible.

They thought there would be one on its way to Dave at some point ( and that contrary to rumours, Dave hadn't pissed them off) , though reading between the lines it seems Tek place a LOT less importance on social media type stuff than Keysight. (I did offer to do a teardown vid but they didn't seem too interested..!).

There is new firmware coming that will add some things that are currently missing, including roll mode, power analysis, segmented memory, and another major thing the guy couldn't immediately recall.

 I mentioned the issue that someone brought up in the RTB2004 thread that colours should be user-changeable to help colourblind users & they  thought it was a good point & would feed it back.
There were a few small things that I remarked as I was going along that the RTB UI did that they could benefit from copying.

They are planning a "headless" version, like the old Agilent L series rackmount units.
No indication that the ASICs will find their way into lower-end models any time soon ( but no denial either so hard to say).
Unlimited number of decodes ( which suggests all in software, as does some slowdown when enabled)  UART up to 10Mbaud, and has options for 10/100Ethernet and USB up to HS 480M ( all extra cost of course).
It does UART packet framing (like the RTB will once they fix the bug) , but does it wrong. You can only frame by terminating character ( and only certain chars - from memory  00 0a 0d  and ff, not any user defineable char), and no option to  frame by timeout. The way that sparse data bytes are presented is poor to the point of being barely useable in some situations as the length of the "envelope" for the text is limited to the length of the data onscreen. e.g. if you have two 9600 baud bytes(~1ms long) , 20mS apart onscreen, you can't read the decode data for either of them ( unless you use packet mode, which loses all timing relationship between the waveform and the decode data) .

Decodes are handled like the RTB - as an extra trace with size and position adjust, but seem to only be 2 size settings that are barely different - nothing like as good as how the RTB will show vertical columns of digits to get more on the screen.
 
Probably the worst thing was that at slow timebase speeds, it does NOT do continuous screen update ( i.e the new data sweeping across the screen) , it just updates the whole screen at the end of each sweep. I spent a long time failing to make their guys understand why this was a problem - I think I'll send them a video to illustrate the point. I did persuade them that roll mode, while useful for some slow waveforms would not solve this for cases where you have a slow event that you want to see what happens after a trigger without waiting however many seconds it takes to fill the screen.

When I mentioned the issue Dave raised of losing analogue channels when you want digital, they claimed their pricing was such that their 6ch unit with 2x digital probes was comparable to competitors' 4+16ch offerings. I've not checked that but am skeptical.

[lukier]

The more I read about this scope the more it looks to me like Tek tried to make a LeCroy scope. From the video mentioning the simulation mode for development to the very similar UI (except for sluggishness and stupid recycle bin idea), to the 8 channel idea (also Yokogawa) and to the PC taking some of the processing (decoding). But the probe interface is quite novel to be honest and I don't mind that as much as Dave does, it is quite logical from DSO design point of view. AFAIR on many scopes one looses something (memory/sampling rates) when MSO is enabled.

On the channel colors, well, LeCroy has it

[JPortici]

Probably the worst thing was that at slow timebase speeds, it does NOT do continuous screen update ( i.e the new data sweeping across the screen) , it just updates the whole screen at the end of each sweep. I spent a long time failing to make their guys understand why this was a problem - I think I'll send them a video to illustrate the point. I did persuade them that roll mode, while useful for some slow waveforms would not solve this for cases where you have a slow event that you want to see what happens after a trigger without waiting however many seconds it takes to fill the screen.

Ah, "Scan" mode. I love using the tek TPS for very slow stuff because it has "Scan" at low timebases. At 100ms/div the 10div screen show exactly one second and the way it updates feels so natural that i copied it in some of our software
i find it unbeliveable that tek was the only one to implement this alternative to roll mode.. and even more that they aren't doing it anymore.

I like it a lot more than "Roll" because on roll the picture is always moving, instead with "scan" it's like a radar, the picture is fixed but for the scanline

RE: Colors, i can change trace colors on my 1989 lecroy too

[mikeselectricstuff]

Probably the worst thing was that at slow timebase speeds, it does NOT do continuous screen update ( i.e the new data sweeping across the screen) , it just updates the whole screen at the end of each sweep. I spent a long time failing to make their guys understand why this was a problem - I think I'll send them a video to illustrate the point. I did persuade them that roll mode, while useful for some slow waveforms would not solve this for cases where you have a slow event that you want to see what happens after a trigger without waiting however many seconds it takes to fill the screen.

Ah, "Scan" mode. I love using the tek TPS for very slow stuff because it has "Scan" at low timebases. At 100ms/div the 10div screen show exactly one second and the way it updates feels so natural that i copied it in some of our software
i find it unbeliveable that tek was the only one to implement this alternative to roll mode.. and even more that they aren't doing it anymore.

I like it a lot more than "Roll" because on roll the picture is always moving, instead with "scan" it's like a radar, the picture is fixed but for the scanline

I'd regard any scope that doesn't do this as simpy broken.
I initially thought my MSOX3104T didn't , but then noticed I had decodes enabled, which disables it.(though I think it transitions about one timebase stop away from where it should)
Don't recall what the RTB does but would be disappointed if it doesn't
For triggered events on slow timebase, not updating in realtime makes it borderline useless for some applications
At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

[tautech]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Siglent's do this ^ but you can opt out of Roll mode to just a slow timebase setting instead.

[Hydron]

The RTB2k immediately starts updating post-trigger waveform on the screen once it triggers. While it does so, everything pre-trigger on the screen is from the previous acquisition until the current acquisition is finished, at which point it is replaced by the pre-trigger record.
This is much better than waiting for the entire triggered acquisition to finish before showing anything, but it isn't perfect having to wait before the pre-trigger record is updated - maybe there's a memory architecture limitation that forces this behaviour (i.e. it can't pull the earlier data from memory while busy writing the post-trigger data to it).

Mike - any opinion of the worthwhile-ness of EDS this year? I was considering going tomorrow, but the timing isn't great with other commitments, so won't unless its really worth it. Did you get a chance to talk to the R&S guys as well? (Apologies for veering a little off-topic!)

edit: RTB starts the scan mode thing at 50ms/div and slower. Seems like a reasonable choice.

[mikeselectricstuff]

Mike - any opinion of the worthwhile-ness of EDS this year? I was considering going tomorrow, but the timing isn't great with other commitments, so wouldn't unless it was really worth it. Did you get a chance to talk to the R&S guys as well? (Apologies for veering a little off-topic!)

Definitely worth it for me, unlike Southern which seems pretty much dead now for Electronics.
Yes I had a long chat with the R&S guy, who I'd not met before.

[rsjsouza]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Siglent's do this ^ but you can opt out of Roll mode to just a slow timebase setting instead.

I am almost 100% sure my DS4014 does this, but since I don't do many slow scans I can't tell if this can be overriden.

[EEVblog]

There is new firmware coming that will add some things that are currently missing, including roll mode, power analysis, segmented memory, and another major thing the guy couldn't immediately recall.

They released their "game changing" flagship scope without roll mode and segmented memory? 

[mikeselectricstuff]

There is new firmware coming that will add some things that are currently missing, including roll mode, power analysis, segmented memory, and another major thing the guy couldn't immediately recall.

They released their "game changing" flagship scope without roll mode and segmented memory? 

He said that was totally planned - they accepted that they couldn't get it all in the first release, and they had to get it out the door, so they obviously focussed on the "sexy" stuff.

[David Hess]

They released their "game changing" flagship scope without roll mode and segmented memory? 

Dude, roll mode is hard.  I had to buy a DSO old enough to drink to get both roll and scan mode.

[mikeselectricstuff]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Siglent's do this ^ but you can opt out of Roll mode to just a slow timebase setting instead.

But does  it then do a live screen update, or only refresh at the end of the sweep?

[tautech]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Siglent's do this ^ but you can opt out of Roll mode to just a slow timebase setting instead.

But does  it then do a live screen update, or only refresh at the end of the sweep?

Roll is swept live from the right edge of the display while slow timebase refreshes at the end of the sweep to display a stationary raster just like a faster speed waveform would display.
Slow timebase settings (not Roll) offer more/better Trigger options, eg. Normal and Single trigger, that's not available in Roll mode.
Each user will have a preference for what they like or need.

[mikeselectricstuff]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Siglent's do this ^ but you can opt out of Roll mode to just a slow timebase setting instead.

But does  it then do a live screen update, or only refresh at the end of the sweep?

Roll is swept live from the right edge of the display while slow timebase refreshes at the end of the sweep to display a stationary raster just like a faster speed waveform would display.
Slow timebase settings (not Roll) offer more/better Trigger options, eg. Normal and Single trigger, that's not available in Roll mode.
Each user will have a preference for what they like or need.

Roll has its uses, and is a seperate thing to how the scope behaves at slow timebase settings in triggered mode.
Not live-updating on a triggered slow capture is always less useful than updating on a full screen.
 Having to wait for the full screen to see what's going on is always less useful, and gets increasingly annoying as the timebase period increases, to the point of being unuseable.

If a scope's architecture doesn't allow live update at slow timebaes, a reasonable alternative would be to implement a triggered roll mode, with options to start on trigger and display a screenful, or continuously roll until a trigger and stop.  Of course you can't expect a Chinese manufacturer to have the imagination to implement something like that.

[tautech]

If a scope's architecture doesn't allow live update at slow timebaes, a reasonable alternative would be to implement a triggered roll mode, with options to start on trigger and display a screenful, or continuously roll until a trigger and stop.  Of course you can't expect a Chinese manufacturer to have the imagination to implement something like that.

Really ?
 

[mikeselectricstuff]

If a scope's architecture doesn't allow live update at slow timebaes, a reasonable alternative would be to implement a triggered roll mode, with options to start on trigger and display a screenful, or continuously roll until a trigger and stop.  Of course you can't expect a Chinese manufacturer to have the imagination to implement something like that.

Really ?
 

Absolutely.
I'm struggling to think of a single example of a Chinese test equipment manufacturer coming up with a new useful feature that isn't just copy of a feature on a big-brand device.
 The Chinese education system places no value on creativity & imagination, which is a big part of why we very rarely see anything truly innovative coming from China. 

[rsjsouza]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Siglent's do this ^ but you can opt out of Roll mode to just a slow timebase setting instead.

But does  it then do a live screen update, or only refresh at the end of the sweep?

Roll is swept live from the right edge of the display while slow timebase refreshes at the end of the sweep to display a stationary raster just like a faster speed waveform would display.
Slow timebase settings (not Roll) offer more/better Trigger options, eg. Normal and Single trigger, that's not available in Roll mode.
Each user will have a preference for what they like or need.

Roll has its uses, and is a seperate thing to how the scope behaves at slow timebase settings in triggered mode.
Not live-updating on a triggered slow capture is always less useful than updating on a full screen.
 Having to wait for the full screen to see what's going on is always less useful, and gets increasingly annoying as the timebase period increases, to the point of being unuseable.

If a scope's architecture doesn't allow live update at slow timebaes, a reasonable alternative would be to implement a triggered roll mode, with options to start on trigger and display a screenful, or continuously roll until a trigger and stop.  Of course you can't expect a Chinese manufacturer to have the imagination to implement something like that.

The DS4014 does allow to select roll or Y-T modes in slow timebases (roll mode goes down to 200ms/div).

In roll mode it updates the screen at the arbitrary Horizontal Reference point - by default at the center of the screen but it can be set anywhere else. Set it to the leftmost X position and the image is updated through the entire time; set it to the rightmost and the image will be updated at the completion of the sweep. To me this is a proper implementation, don't you agree?

[mikeselectricstuff]

At least one Chinese scope (don't recall which) forces roll mode below a certain timebase setting, which isn't especially useful.

Micsig Tbook

[fcb]

I had a short play with the Tek 58 at EDS yesterday. Mike is right, that thing is huge (which isn't actually a problem).  It didn't feel too intuitive, the knobs felt quite crammed and I kept grabbing the wrong knob for the time-base, the pinch-zoom would scare me as I've often got a probe in each hand & i'd hate to scratch that huge, expensive LCD.

On my bench, the 'scopes are on a shelf built in, the size of the plugin module could be an issue and result in having to push the 'scope back - which would be pity as that nice big display is great for my aging eyes.

Price was fairly massive, but if it's use all day, then it's only the price of a very ordinary car in the UK.  I quite liked the feel of the LeCroy 'scopes, touch interface felt a bit cramped - Tek definatlly looked like a LeCroy+ copy.  Look forward to getting both in the lab over to evaluate and then buying a Keysight - probably.

What was a pleasant surprise was Pico's mid-level headless scopes when connected to large display.